High-throughput identification of prefusion-stabilizing mutations in SARS-CoV-2 spike

Abstract

Designing prefusion-stabilized SARS-CoV-2 spike is critical for the effectiveness of COVID-19 vaccines. All COVID-19 vaccines in the US encode spike with K986P/V987P mutations to stabilize its prefusion conformation. However, contemporary methods on engineering prefusion-stabilized spike immunogens involve tedious experimental work and heavily rely on structural information. Here, we establish a systematic and unbiased method of identifying mutations that concomitantly improve expression and stabilize the prefusion conformation of the SARS-CoV-2 spike. Our method integrates a fluorescence-based fusion assay, mammalian cell display technology, and deep mutational scanning. As a proof-of-concept, we apply this method to a region in the S2 domain that includes the first heptad repeat and central helix. Our results reveal that besides K986P and V987P, several mutations simultaneously improve expression and significantly lower the fusogenicity of the spike. As prefusion stabilization is a common challenge for viral immunogen design, this work will help accelerate vaccine development against different viruses.

Evidence of antigenic drift in the fusion machinery core of SARS-CoV-2 spike

“…Our study here suggests a similar phenomenon in the relatively conserved core fusion S2 antibodies include those that abolish fusion activity (e.g. T961F and Q1005R) 24 and enhance fusion activity (e.g. D950N) 27 .…”

“…Given that IGHV1-69/IGKV3-11 S2 antibodies exhibited in vivo protection activity, we were interested in whether mutations in S2 could influence their binding activity. To systematically identify mutations on the S protein that affect binding to IGHV1-69/IGKV3-11 S2 antibodies, we performed a deep mutational scanning (DMS) experiment using our previously constructed S mutant library, which contained all possible amino acid mutations from residues 883 to 1034, spanning HR1 and CH in the S2 domain 24 . Briefly, this mutant library was displayed on human embryonic kidney 293T (HEK293T) landing pad cells.…”

“…Since our deep mutational scanning experiments relied on cell surface display, mutations that lowered the cell-surface expression level would result in a lower binding score even if it did not affect the antibody binding affinity. Therefore, we compared the binding scores of individual mutations to their cell-surface expression levels which were previously quantified using the RBD antibody CC12.3 24,25 . As expected, there was a mild correlation between binding score and expression score, which is a proxy for cell-surface expression level 24 (Pearson correlation coefficient of 0.48 to 0.49, Figure 2 ).…”

“…HEK293T landing pad cells expressing the S2 HR1/CH mutant library of S as constructed previously 24 were harvested and centrifuged at 300 ´ g for 5 min at 4 °C. Supernatant was discarded.…”

Immunization with V987H-stabilized Spike glycoprotein protects K18-hACE2 and golden Syrian hamster upon SARS-CoV-2 infection.

“…In addition, Riley and coworkers cross-linked the S2 subunit with different regions of the S1 subunit by introducing disulphide bridges, which also limited the motility of the RBD 29 . Besides these structure-guided mutation selection strategies, Tan et al developed a high-throughput methods for the identi cation of pre-fusion stabilizing mutations within the heptad repeat 1 and central helix regions of the S2 subunit 30 . However, whether all these S protein variants show an improved immunogenicity or protective capabilities compared to the S-2P protein remains unknown.…”